AI agents
Agents access approved information and tools, handle defined tasks and hand results to people or downstream systems.
00
AI · Workflows · System Integration
You bring the data and systems. I build the solutions for you.
I build AI agents, automated workflows and internal applications that work with your existing environment.
No obligation — tell me your systems and processes, and I will sketch a clear technical starting point
01 — Solution forms
Three kinds of solutions for concrete operational tasks
The technical form follows the task. What matters is not whether the result is an agent, a workflow or an application, but whether the operational process works better as a result.
Automated workflows
Workflows connect data, business rules, deadlines, approvals and human decisions into a traceable process.
Internal applications
Focused tools cover tasks for which standard systems are too generic and Excel or email are no longer enough.
02 — How it connects
A working solution connects more than tools alone
A new assistant, workflow or internal application achieves little if information is still transferred manually, rules remain outside the system or people have to switch between several interfaces.
That is why the solution is built along the actual workflow.
Input and use
People work through forms, portals, dashboards, chats or focused business interfaces.
Process and logic
Validations, calculations, approvals, deadlines and AI-supported processing determine what happens next.
Data and knowledge
Documents, emails, SharePoint, databases and business systems supply the information the task requires.
Systems and interfaces
Microsoft 365, Power Automate, n8n, APIs, SQL and existing applications are connected where handovers still happen manually today.
Control and operation
Permissions, logging, quality checks and designed error paths ensure the solution can be operated in a traceable way.
03 — Scope
Scope follows the task
Some initiatives are clearly bounded; others touch several areas and systems. What matters is choosing a scope that lets the solution be delivered as a complete business fit and a technically viable system — not artificially as small or as large as possible.
Clarify goal and boundaries
At the outset we define which task the solution should take on, which areas are affected and how a successful outcome will be recognised.
Make dependencies visible
Data sources, interfaces, existing systems, roles and external parties are considered early.
Structure delivery sensibly
Larger initiatives are broken into traceable work packages without losing sight of the shared target picture.
Test early under real conditions
Important functions, integrations and assumptions are checked in time with realistic data and the people who will use the system later.
Deliver with control
Depending on its size, the solution is introduced step by step or as one coherent initiative — with clear ownership and defined handovers.
Not more technology. The right connection.
A good solution does not automatically replace your existing environment. It uses what already works, connects what is separate today and adds only what the concrete task still needs.
04 — Examples
What concrete solutions can look like
The exact technical form follows from the task, the systems already in place and the requirements for later operation. Typical initiatives connect several building blocks into one end-to-end workflow.
Prepare incoming requests automatically
Outlook
Power Automateor
n8n
OpenAI
SharePoint
Emails or forms are captured, classified and matched against existing information.
AI extracts relevant details, creates a summary or prepares a reply. Rules decide which cases continue automatically and where a person must review or approve.
Make internal documents usable with intent
- SharePoint
OneDrive- OpenAIor
Claude - vector search
An AI agent searches approved policies, manuals, project documents or other internal sources and answers concrete questions with traceable references.
Permissions, source freshness and how uncertain results are handled are built into the solution.
Guide requests, checks and approvals
Power Appsorweb app- Power Automate
- SharePoint
SQL
People are guided through the required inputs and decisions. The system checks entries, calculates values, handles exceptions and includes the mandated approval stages.
Deadlines, follow-up questions and escalations remain traceable inside the process.
Map a specific business task into an application
Vue
JavaScript- SQLor
Supabase - APIs
For tasks that do not belong in a standard system and cannot sensibly live in Excel forever, a focused internal application is built.
It connects a fitting interface with existing business logic, data and the necessary integrations.
05 — Technology
The right tool for each part of the solution
Not every function should be built with AI. Not every process needs custom development. And not every existing platform has to be replaced.
Microsoft 365 and Power Platform
For document-based processes, inputs, approvals, reporting and integrations inside an existing Microsoft environment.
- SharePoint
- Outlook
Teams- OneDrive
Excel
Forms- Power Apps
- Power Automate
Power BI
Azure AD- Lists
- Graph
Automation and system integration
For cross-system processes, CRM connections, event processing and linking different business systems.
- n8n
Make
Zapier
FireStart
Salesforce
HubSpot
Azure Service Bus- APIs
- webhooks
AI agents and language models
For processing language, documents and unstructured information, as well as clearly bounded tasks with access to internal tools.
- OpenAI
Azure OpenAI
Copilot- Copilot Studio
- Claude
Gemini
LangChain- Whisper
Data and backend
For structured storage, queries, caching and connecting existing data stores to applications and automations.
- SQL
PostgreSQL- Supabase
Redis
Focused custom development
For specific business applications, tailored user guidance and requirements that standard tools cannot cover sensibly.
- Vue
Nuxt- JavaScript
Node.js
Tailwind
Docker
Cursor
GitHub Copilot
06 — Operational readiness
It is not finished when the demo works
A solution must not only handle the intended process. It also has to cope with incomplete inputs, exceptions, technical failures and changing requirements.
That is why quality, governance and the prerequisites for later operation are built as part of delivery — not added afterwards.
Integrated
- Data
- interfaces
- permissions
The solution reaches the intended sources, passes information reliably and respects existing roles, access rights and data-use requirements.
Safeguarded
- Rules
- tests
- quality controls
Business logic, checks, relevant exceptions and designed error handling are implemented in a traceable way and tested under realistic conditions.
Traceable
- Logging
- documentation
- decisions
Important cases, approvals and system decisions are documented where quality, control and operational accountability require it.
Observable
- Monitoring
- errors
- quality
Errors and relevant deviations become visible. Owners can understand what the system did, how results were produced and where intervention is needed.
Adoptable
- Interface
- training
- use
The solution is designed and introduced so people understand how to work with it, which results need review and what to do when something is unclear.
Ownable
- Documentation
- ownership
- changes
System, logic and the necessary operating steps are documented. Ownership for operation, quality control, adjustments and further development is clarified.
07 — Contact
Let’s define the first deliverable scope
You describe the existing process, the systems involved and the outcome you want.
Together we bound a first scope that is complete for the business, technically realistic and genuinely usable in day-to-day work.
We also clarify which data, interfaces, permissions and operational responsibilities have to be taken into account.